gitweb: Use git-diff-tree or git-diff patch output for blobdiff
[gitweb.git] / gitweb / gitweb.perl
index 93676857344858ba84c1cbb82351d6c2ecf8a105..0fa35f336994df9756d372c2c89169d0875091c9 100755 (executable)
@@ -211,6 +211,13 @@ sub feature_snapshot {
        }
 }
 
+our $hash_parent_base = $cgi->param('hpb');
+if (defined $hash_parent_base) {
+       if (!validate_input($hash_parent_base)) {
+               die_error(undef, "Invalid hash parent base parameter");
+       }
+}
+
 our $page = $cgi->param('pg');
 if (defined $page) {
        if ($page =~ m/[^0-9]$/) {
@@ -270,13 +277,14 @@ (%)
        my %params = @_;
 
        my @mapping = (
-               action => "a",
                project => "p",
+               action => "a",
                file_name => "f",
                file_parent => "fp",
                hash => "h",
                hash_parent => "hp",
                hash_base => "hb",
+               hash_parent_base => "hpb",
                page => "pg",
                searchtext => "s",
        );
@@ -446,7 +454,13 @@ sub mode_str {
 
 # convert file mode in octal to file type string
 sub file_type {
-       my $mode = oct shift;
+       my $mode = shift;
+
+       if ($mode !~ m/^[0-7]+$/) {
+               return $mode;
+       } else {
+               $mode = oct $mode;
+       }
 
        if (S_ISDIR($mode & S_IFMT)) {
                return "directory";
@@ -538,8 +552,7 @@ sub format_diff_line {
        } elsif ($char eq "@") {
                $diff_class = " chunk_header";
        } elsif ($char eq "\\") {
-               # skip errors (incomplete lines)
-               return "";
+               $diff_class = " incomplete";
        }
        $line = untabify($line);
        return "<div class=\"diff$diff_class\">" . esc_html($line) . "</div>\n";
@@ -611,6 +624,26 @@ sub git_get_hash_by_path {
        return $3;
 }
 
+# converts symbolic name to hash
+sub git_to_hash {
+       my @params = @_;
+       return undef unless @params;
+
+       open my $fd, "-|", $GIT, "rev-parse", @params
+               or return undef;
+       my @hashes = map { chomp; $_ } <$fd>;
+       close $fd;
+
+       if (wantarray) {
+               return @hashes;
+       } elsif (scalar(@hashes) == 1) {
+               # single hash
+               return $hashes[0];
+       } else {
+               return \@hashes;
+       }
+}
+
 ## ......................................................................
 ## git utility functions, directly accessing git repository
 
@@ -750,6 +783,73 @@ sub git_get_references {
        return \%refs;
 }
 
+sub git_get_following_references {
+       my $hash = shift || return undef;
+       my $type = shift;
+       my $base = shift || $hash_base || "HEAD";
+
+       my $refs = git_get_references($type);
+       open my $fd, "-|", $GIT, "rev-list", $base
+               or return undef;
+       my @commits = map { chomp; $_ } <$fd>;
+       close $fd
+               or return undef;
+
+       my @reflist;
+       my $lastref;
+
+       foreach my $commit (@commits) {
+               foreach my $ref (@{$refs->{$commit}}) {
+                       $lastref = $ref;
+                       push @reflist, $lastref;
+               }
+               if ($commit eq $hash) {
+                       last;
+               }
+       }
+
+       return wantarray ? @reflist : $lastref;
+}
+
+sub git_get_preceding_references {
+       my $hash = shift || return undef;
+       my $type = shift;
+
+       my $refs = git_get_references($type);
+       open my $fd, "-|", $GIT, "rev-list", $hash
+               or return undef;
+       my @commits = map { chomp; $_ } <$fd>;
+       close $fd
+               or return undef;
+
+       my @reflist;
+
+       foreach my $commit (@commits) {
+               foreach my $ref (@{$refs->{$commit}}) {
+                       return $ref unless wantarray;
+                       push @reflist, $ref;
+               }
+       }
+
+       return @reflist;
+}
+
+sub git_get_rev_name_tags {
+       my $hash = shift || return undef;
+
+       open my $fd, "-|", $GIT, "name-rev", "--tags", $hash
+               or return;
+       my $name_rev = <$fd>;
+       close $fd;
+
+       if ($name_rev =~ m|^$hash tags/(.*)$|) {
+               return $1;
+       } else {
+               # catches also '$hash undefined' output
+               return undef;
+       }
+}
+
 ## ----------------------------------------------------------------------
 ## parse to hash functions
 
@@ -1477,8 +1577,10 @@ sub git_difftree_body {
                        }
                        print "<td>";
                        if ($diff{'to_id'} ne $diff{'from_id'}) { # modified
-                               print $cgi->a({-href => href(action=>"blobdiff", hash=>$diff{'to_id'}, hash_parent=>$diff{'from_id'},
-                                                            hash_base=>$hash, file_name=>$diff{'file'}),
+                               print $cgi->a({-href => href(action=>"blobdiff",
+                                                            hash=>$diff{'to_id'}, hash_parent=>$diff{'from_id'},
+                                                            hash_base=>$hash, hash_parent_base=>$parent,
+                                                            file_name=>$diff{'file'}),
                                              -class => "list"}, esc_html($diff{'file'}));
                        } else { # only mode changed
                                print $cgi->a({-href => href(action=>"blob", hash=>$diff{'to_id'},
@@ -1493,8 +1595,10 @@ sub git_difftree_body {
                                        "blob");
                        if ($diff{'to_id'} ne $diff{'from_id'}) { # modified
                                print " | " .
-                                       $cgi->a({-href => href(action=>"blobdiff", hash=>$diff{'to_id'}, hash_parent=>$diff{'from_id'},
-                                                              hash_base=>$hash, file_name=>$diff{'file'})},
+                                       $cgi->a({-href => href(action=>"blobdiff",
+                                                              hash=>$diff{'to_id'}, hash_parent=>$diff{'from_id'},
+                                                              hash_base=>$hash, hash_parent_base=>$parent,
+                                                              file_name=>$diff{'file'})},
                                                "diff");
                        }
                        print " | " .
@@ -1526,8 +1630,9 @@ sub git_difftree_body {
                                      "blob");
                        if ($diff{'to_id'} ne $diff{'from_id'}) {
                                print " | " .
-                                       $cgi->a({-href => href(action=>"blobdiff", hash_base=>$hash,
+                                       $cgi->a({-href => href(action=>"blobdiff",
                                                               hash=>$diff{'to_id'}, hash_parent=>$diff{'from_id'},
+                                                              hash_base=>$hash, hash_parent_base=>$parent,
                                                               file_name=>$diff{'to_file'}, file_parent=>$diff{'from_file'})},
                                                "diff");
                        }
@@ -1540,16 +1645,18 @@ sub git_difftree_body {
 }
 
 sub git_patchset_body {
-       my ($patchset, $difftree, $hash, $hash_parent) = @_;
+       my ($fd, $difftree, $hash, $hash_parent) = @_;
 
        my $patch_idx = 0;
        my $in_header = 0;
        my $patch_found = 0;
-       my %diffinfo;
+       my $diffinfo;
 
        print "<div class=\"patchset\">\n";
 
-       LINE: foreach my $patch_line (@$patchset) {
+       LINE:
+       while (my $patch_line @$fd>) {
+               chomp $patch_line;
 
                if ($patch_line =~ m/^diff /) { # "git diff" header
                        # beginning of patch (in patchset)
@@ -1562,54 +1669,60 @@ sub git_patchset_body {
                        }
                        print "<div class=\"patch\">\n";
 
-                       %diffinfo = parse_difftree_raw_line($difftree->[$patch_idx++]);
+                       if (ref($difftree->[$patch_idx]) eq "HASH") {
+                               $diffinfo = $difftree->[$patch_idx];
+                       } else {
+                               $diffinfo = parse_difftree_raw_line($difftree->[$patch_idx]);
+                       }
+                       $patch_idx++;
 
                        # for now, no extended header, hence we skip empty patches
                        # companion to  next LINE if $in_header;
-                       if ($diffinfo{'from_id'} eq $diffinfo{'to_id'}) { # no change
+                       if ($diffinfo->{'from_id'} eq $diffinfo->{'to_id'}) { # no change
                                $in_header = 1;
                                next LINE;
                        }
 
-                       if ($diffinfo{'status'} eq "A") { # added
-                               print "<div class=\"diff_info\">" . file_type($diffinfo{'to_mode'}) . ":" .
+                       if ($diffinfo->{'status'} eq "A") { # added
+                               print "<div class=\"diff_info\">" . file_type($diffinfo->{'to_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash,
-                                                            hash=>$diffinfo{'to_id'}, file_name=>$diffinfo{'file'})},
-                                             $diffinfo{'to_id'}) . "(new)" .
+                                                            hash=>$diffinfo->{'to_id'}, file_name=>$diffinfo->{'file'})},
+                                             $diffinfo->{'to_id'}) . "(new)" .
                                      "</div>\n"; # class="diff_info"
 
-                       } elsif ($diffinfo{'status'} eq "D") { # deleted
-                               print "<div class=\"diff_info\">" . file_type($diffinfo{'from_mode'}) . ":" .
+                       } elsif ($diffinfo->{'status'} eq "D") { # deleted
+                               print "<div class=\"diff_info\">" . file_type($diffinfo->{'from_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash_parent,
-                                                            hash=>$diffinfo{'from_id'}, file_name=>$diffinfo{'file'})},
-                                             $diffinfo{'from_id'}) . "(deleted)" .
+                                                            hash=>$diffinfo->{'from_id'}, file_name=>$diffinfo->{'file'})},
+                                             $diffinfo->{'from_id'}) . "(deleted)" .
                                      "</div>\n"; # class="diff_info"
 
-                       } elsif ($diffinfo{'status'} eq "R" || # renamed
-                                $diffinfo{'status'} eq "C") { # copied
+                       } elsif ($diffinfo->{'status'} eq "R" || # renamed
+                                $diffinfo->{'status'} eq "C" || # copied
+                                $diffinfo->{'status'} eq "2") { # with two filenames (from git_blobdiff)
                                print "<div class=\"diff_info\">" .
-                                     file_type($diffinfo{'from_mode'}) . ":" .
+                                     file_type($diffinfo->{'from_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash_parent,
-                                                            hash=>$diffinfo{'from_id'}, file_name=>$diffinfo{'from_file'})},
-                                             $diffinfo{'from_id'}) .
+                                                            hash=>$diffinfo->{'from_id'}, file_name=>$diffinfo->{'from_file'})},
+                                             $diffinfo->{'from_id'}) .
                                      " -> " .
-                                     file_type($diffinfo{'to_mode'}) . ":" .
+                                     file_type($diffinfo->{'to_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash,
-                                                            hash=>$diffinfo{'to_id'}, file_name=>$diffinfo{'to_file'})},
-                                             $diffinfo{'to_id'});
+                                                            hash=>$diffinfo->{'to_id'}, file_name=>$diffinfo->{'to_file'})},
+                                             $diffinfo->{'to_id'});
                                print "</div>\n"; # class="diff_info"
 
                        } else { # modified, mode changed, ...
                                print "<div class=\"diff_info\">" .
-                                     file_type($diffinfo{'from_mode'}) . ":" .
+                                     file_type($diffinfo->{'from_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash_parent,
-                                                            hash=>$diffinfo{'from_id'}, file_name=>$diffinfo{'file'})},
-                                             $diffinfo{'from_id'}) .
+                                                            hash=>$diffinfo->{'from_id'}, file_name=>$diffinfo->{'file'})},
+                                             $diffinfo->{'from_id'}) .
                                      " -> " .
-                                     file_type($diffinfo{'to_mode'}) . ":" .
+                                     file_type($diffinfo->{'to_mode'}) . ":" .
                                      $cgi->a({-href => href(action=>"blob", hash_base=>$hash,
-                                                            hash=>$diffinfo{'to_id'}, file_name=>$diffinfo{'file'})},
-                                             $diffinfo{'to_id'});
+                                                            hash=>$diffinfo->{'to_id'}, file_name=>$diffinfo->{'file'})},
+                                             $diffinfo->{'to_id'});
                                print "</div>\n"; # class="diff_info"
                        }
 
@@ -1620,8 +1733,30 @@ sub git_patchset_body {
 
 
                if ($in_header && $patch_line =~ m/^---/) {
-                       #print "</div>\n"
+                       #print "</div>\n"; # class="diff extended_header"
                        $in_header = 0;
+
+                       my $file = $diffinfo->{'from_file'};
+                       $file  ||= $diffinfo->{'file'};
+                       $file = $cgi->a({-href => href(action=>"blob", hash_base=>$hash_parent,
+                                                      hash=>$diffinfo->{'from_id'}, file_name=>$file),
+                                       -class => "list"}, esc_html($file));
+                       $patch_line =~ s|a/.*$|a/$file|g;
+                       print "<div class=\"diff from_file\">$patch_line</div>\n";
+
+                       $patch_line = <$fd>;
+                       chomp $patch_line;
+
+                       #$patch_line =~ m/^+++/;
+                       $file    = $diffinfo->{'to_file'};
+                       $file  ||= $diffinfo->{'file'};
+                       $file = $cgi->a({-href => href(action=>"blob", hash_base=>$hash,
+                                                      hash=>$diffinfo->{'to_id'}, file_name=>$file),
+                                       -class => "list"}, esc_html($file));
+                       $patch_line =~ s|b/.*|b/$file|g;
+                       print "<div class=\"diff to_file\">$patch_line</div>\n";
+
+                       next LINE;
                }
                next LINE if $in_header;
 
@@ -1725,8 +1860,10 @@ sub git_history_body {
                        if (defined $blob_current && defined $blob_parent &&
                                        $blob_current ne $blob_parent) {
                                print " | " .
-                                       $cgi->a({-href => href(action=>"blobdiff", hash=>$blob_current, hash_parent=>$blob_parent,
-                                                              hash_base=>$commit, file_name=>$file_name)},
+                                       $cgi->a({-href => href(action=>"blobdiff",
+                                                              hash=>$blob_current, hash_parent=>$blob_parent,
+                                                              hash_base=>$hash_base, hash_parent_base=>$commit,
+                                                              file_name=>$file_name)},
                                                "diff to current");
                        }
                }
@@ -2678,34 +2815,122 @@ sub git_commit {
 }
 
 sub git_blobdiff {
-       mkdir($git_temp, 0700);
+       my $fd;
+       my @difftree;
+       my %diffinfo;
+
+       # preparing $fd and %diffinfo for git_patchset_body
+       # new style URI
+       if (defined $hash_base && defined $hash_parent_base) {
+               if (defined $file_name) {
+                       # read raw output
+                       open $fd, "-|", $GIT, "diff-tree", '-r', '-M', '-C', $hash_parent_base, $hash_base,
+                               "--", $file_name
+                               or die_error(undef, "Open git-diff-tree failed");
+                       @difftree = map { chomp; $_ } <$fd>;
+                       close $fd
+                               or die_error(undef, "Reading git-diff-tree failed");
+                       @difftree
+                               or die_error('404 Not Found', "Blob diff not found");
+
+               } elsif (defined $hash) { # try to find filename from $hash
+                       if ($hash !~ /[0-9a-fA-F]{40}/) {
+                               $hash = git_to_hash($hash);
+                       }
+
+                       # read filtered raw output
+                       open $fd, "-|", $GIT, "diff-tree", '-r', '-M', '-C', $hash_parent_base, $hash_base
+                               or die_error(undef, "Open git-diff-tree failed");
+                       @difftree =
+                               # ':100644 100644 03b21826... 3b93d5e7... M     ls-files.c'
+                               # $hash == to_id
+                               grep { /^:[0-7]{6} [0-7]{6} [0-9a-fA-F]{40} $hash/ }
+                               map { chomp; $_ } <$fd>;
+                       close $fd
+                               or die_error(undef, "Reading git-diff-tree failed");
+                       @difftree
+                               or die_error('404 Not Found', "Blob diff not found");
+
+               } else {
+                       die_error('404 Not Found', "Missing one of the blob diff parameters");
+               }
+
+               if (@difftree > 1) {
+                       die_error('404 Not Found', "Ambiguous blob diff specification");
+               }
+
+               %diffinfo = parse_difftree_raw_line($difftree[0]);
+               $file_parent ||= $diffinfo{'from_file'} || $file_name || $diffinfo{'file'};
+               $file_name   ||= $diffinfo{'to_file'}   || $diffinfo{'file'};
+
+               $hash_parent ||= $diffinfo{'from_id'};
+               $hash        ||= $diffinfo{'to_id'};
+
+               # open patch output
+               open $fd, "-|", $GIT, "diff-tree", '-r', '-p', '-M', '-C', $hash_parent_base, $hash_base,
+                       "--", $file_name
+                       or die_error(undef, "Open git-diff-tree failed");
+       }
+
+       # old/legacy style URI
+       if (!%diffinfo && # if new style URI failed
+           defined $hash && defined $hash_parent) {
+               # fake git-diff-tree raw output
+               $diffinfo{'from_mode'} = $diffinfo{'to_mode'} = "blob";
+               $diffinfo{'from_id'} = $hash_parent;
+               $diffinfo{'to_id'}   = $hash;
+               if (defined $file_name) {
+                       if (defined $file_parent) {
+                               $diffinfo{'status'} = '2';
+                               $diffinfo{'from_file'} = $file_parent;
+                               $diffinfo{'to_file'}   = $file_name;
+                       } else { # assume not renamed
+                               $diffinfo{'status'} = '1';
+                               $diffinfo{'from_file'} = $file_name;
+                               $diffinfo{'to_file'}   = $file_name;
+                       }
+               } else { # no filename given
+                       $diffinfo{'status'} = '2';
+                       $diffinfo{'from_file'} = $hash_parent;
+                       $diffinfo{'to_file'}   = $hash;
+               }
+
+               #open $fd, "-|", $GIT, "diff", '-p', $hash_parent, $hash
+               open $fd, "-|", $GIT, "diff", '-p', $hash, $hash_parent
+                       or die_error(undef, "Open git-diff failed");
+       } else  {
+               die_error('404 Not Found', "Missing one of the blob diff parameters")
+                       unless %diffinfo;
+       }
+
+       # header
+       my $formats_nav =
+               $cgi->a({-href => href(action=>"blobdiff_plain",
+                                      hash=>$hash, hash_parent=>$hash_parent,
+                                      hash_base=>$hash_base, hash_parent_base=>$hash_parent_base,
+                                      file_name=>$file_name, file_parent=>$file_parent)},
+                       "plain");
        git_header_html();
        if (defined $hash_base && (my %co = parse_commit($hash_base))) {
-               my $formats_nav =
-                       $cgi->a({-href => href(action=>"blobdiff_plain",
-                                              hash=>$hash, hash_parent=>$hash_parent)},
-                               "plain");
                git_print_page_nav('','', $hash_base,$co{'tree'},$hash_base, $formats_nav);
                git_print_header_div('commit', esc_html($co{'title'}), $hash_base);
        } else {
-               print <<HTML;
-<div class="page_nav"><br/><br/></div>
-<div class="title">$hash vs $hash_parent</div>
-HTML
+               print "<div class=\"page_nav\"><br/>$formats_nav<br/></div>\n";
+               print "<div class=\"title\">$hash vs $hash_parent</div>\n";
        }
-       git_print_page_path($file_name, "blob", $hash_base);
-       print "<div class=\"page_body\">\n" .
-             "<div class=\"diff_info\">blob:" .
-             $cgi->a({-href => href(action=>"blob", hash=>$hash_parent,
-                                    hash_base=>$hash_base, file_name=>($file_parent || $file_name))},
-                     $hash_parent) .
-             " -> blob:" .
-             $cgi->a({-href => href(action=>"blob", hash=>$hash,
-                                    hash_base=>$hash_base, file_name=>$file_name)},
-                     $hash) .
-             "</div>\n";
-       git_diff_print($hash_parent, $file_name || $hash_parent, $hash, $file_name || $hash);
-       print "</div>"; # page_body
+       if (defined $file_name) {
+               git_print_page_path($file_name, "blob", $hash_base);
+       } else {
+               print "<div class=\"page_path\"></div>\n";
+       }
+
+       # patch
+       print "<div class=\"page_body\">\n";
+
+       git_patchset_body($fd, [ \%diffinfo ], $hash_base, $hash_parent_base);
+       close $fd;
+
+       print "</div>\n"; # class="page_body"
        git_footer_html();
 }
 
@@ -2728,7 +2953,6 @@ sub git_commitdiff {
        # read commitdiff
        my $fd;
        my @difftree;
-       my @patchset;
        if ($format eq 'html') {
                open $fd, "-|", $GIT, "diff-tree", '-r', '-M', '-C',
                        "--patch-with-raw", "--full-index", $hash_parent, $hash
@@ -2739,13 +2963,11 @@ sub git_commitdiff {
                        last unless $line;
                        push @difftree, $line;
                }
-               @patchset = map { chomp; $_ } <$fd>;
 
-               close $fd
-                       or die_error(undef, "Reading git-diff-tree failed");
        } elsif ($format eq 'plain') {
                open $fd, "-|", $GIT, "diff-tree", '-r', '-p', '-B', $hash_parent, $hash
                        or die_error(undef, "Open git-diff-tree failed");
+
        } else {
                die_error(undef, "Unknown commitdiff format");
        }
@@ -2775,10 +2997,7 @@ sub git_commitdiff {
 
        } elsif ($format eq 'plain') {
                my $refs = git_get_references("tags");
-               my @tagnames;
-               if (exists $refs->{$hash}) {
-                       @tagnames = map { s|^tags/|| } $refs->{$hash};
-               }
+               my $tagname = git_get_rev_name_tags($hash);
                my $filename = basename($project) . "-$hash.patch";
 
                print $cgi->header(
@@ -2792,10 +3011,9 @@ sub git_commitdiff {
 Date: $ad{'rfc2822'} ($ad{'tz_local'})
 Subject: $co{'title'}
 TEXT
-               foreach my $tag (@tagnames) {
-                       print "X-Git-Tag: $tag\n";
-               }
+               print "X-Git-Tag: $tagname\n" if $tagname;
                print "X-Git-Url: " . $cgi->self_url() . "\n\n";
+
                foreach my $line (@{$co{'comment'}}) {
                        print "$line\n";
                }
@@ -2807,8 +3025,8 @@ sub git_commitdiff {
                #git_difftree_body(\@difftree, $hash, $hash_parent);
                #print "<br/>\n";
 
-               git_patchset_body(\@patchset, \@difftree, $hash, $hash_parent);
-
+               git_patchset_body($fd, \@difftree, $hash, $hash_parent);
+               close $fd;
                print "</div>\n"; # class="page_body"
                git_footer_html();